User authentication for voice-input devices

Abstract

Techniques for authenticating users at devices that interact with the users via voice input. For instance, the described techniques may allow a voice-input device to safely verify the identity of a user by engaging in a back-and-forth conversation. The device or another device coupled thereto may then verify the accuracy of the responses from the user during the conversation, as well as compare an audio signature associated with the user's responses to a pre-stored audio signature associated with the user. By utilizing multiple checks, the described techniques are able to accurately and safely authenticate the user based solely on an audible conversation between the user and the voice-input device.

What is claimed is: 
     
       1. A system comprising:
 one or more processors; and 
 memory storing one or more computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
 determining, in association with a user, one or more first words to be spoken by the user; 
 receiving a first audio signal representing a first utterance spoken by the user; 
 determining one or more first audio characteristics associated with the first audio signal; 
 storing the one or more first audio characteristics in association with a user profile associated with the user; 
 receiving a second audio signal representing a second utterance; 
 processing, using at least one of speech recognition or natural language processing techniques, the second audio signal; 
 determining, based at least partly on processing the second audio signal, that the second utterance includes one or more second words that represent a request for an operation associated with the user profile to be performed; 
 determining one or more second audio characteristics associated with the second audio signal, the one or more second audio characteristics including at least one of pitch, tone, or decibel level associated with the second audio signal; 
 determining that the one or more second audio characteristics are within a threshold similarity with respect to the one or more first audio characteristics; and 
 based at least partly on the request for the operation and the one or more second audio characteristics being within the threshold similarity with respect to the one or more first audio characteristics, causing audio data corresponding to performance of the operation to be output via one or more speakers of a user device associated with the user. 
 
 
     
     
       2. The system as recited in  claim 1 , wherein the operations further comprise:
 determining an audio signature based at least partly on the one or more first audio characteristics; and 
 authenticating the user based at least partly on the audio signature. 
 
     
     
       3. The system as recited in  claim 2 , wherein the operations further comprise determining an identity of the user based at least partly on the one or more second audio characteristics, the one or more second audio characteristics including a frequency associated with the audio signal. 
     
     
       4. The system as recited in  claim 1 , wherein the operations further comprise causing the operation to occur. 
     
     
       5. The system as recited in  claim 1 , wherein the operations further comprise:
 comparing the one or more first audio characteristics to the one or more second audio characteristics; and 
 determining, based at least partly on comparing the one or more first audio characteristics to the one or more second audio characteristics, an identity of the user. 
 
     
     
       6. The system as recited in  claim 1 , wherein the one or more first words are included within a text message sent to the user device. 
     
     
       7. The system as recited in  claim 1 , wherein the operations further comprise:
 receiving the first audio signal from the user device having one or more microphones that captured the first utterance. 
 
     
     
       8. A method comprising:
 receiving an audio signal representing an utterance; 
 processing the audio signal to determine one or more words included in the utterance; 
 determining, based at least partly on processing the audio signal, that the one or more words represent a request for an operation associated with a user profile to be performed; 
 determining one or more first audio characteristics associated with the audio signal, the one or more first audio characteristics including at least one of pitch, tone, or decibel level associated with the audio signal; 
 determining that the one or more first audio characteristics are within a threshold similarity with respect to one or more second audio characteristics previously stored in association with the user profile; and 
 based at least partly on the request for the operation and the one or more first audio characteristics being within the threshold similarity with respect to the one or more second audio characteristics, causing audio data corresponding to performance of the operation to be output via one or more speakers of a user device associated with the user profile. 
 
     
     
       9. The method as recited in  claim 8 , further comprising:
 determining an audio signature based at least partly on the one or more second audio characteristics; and 
 authenticating a user associated with the user profile based at least partly on the audio signature. 
 
     
     
       10. The method as recited in  claim 9 , further comprising determining, based at least partly on the one or more first audio characteristics, an identity of a user associated with the user profile. 
     
     
       11. The method as recited in  claim 9 , wherein the one or more second audio characteristics include a frequency that is identifiable from the audio signal. 
     
     
       12. The method as recited in  claim 8 , further comprising:
 maintaining the one or more second audio characteristics previously stored in association with the user profile; 
 comparing the one or more second audio characteristics to the one or more first audio characteristics; and 
 determining, based at least partly on comparing the one or more second audio characteristics to the one or more first audio characteristics, an identity of the user. 
 
     
     
       13. The method as recited in  claim 8 , further comprising requesting that a user associated with the user profile audibly output one or more second words included within a text message sent to the user device. 
     
     
       14. The method as recited in  claim 8 , further comprising:
 causing, via the one or more speakers of the user device that captured the utterance, output of an audible request for a user associated with the use profile to audibly utter the one or more second words; and 
 determining that the utterance corresponds to the one or more words. 
 
     
     
       15. A method comprising:
 receiving an audio signal representing an utterance, the utterance being captured by one or more microphones of a user device associated with a user profile; 
 processing the audio signal to determine that the utterance includes a request for an operation associated with the user profile to be performed; 
 determining one or more first audio characteristics associated with the audio signal, the one or more first audio characteristics including at least one of pitch, tone, or decibel level associated with the audio signal; 
 determining that the one or more first audio characteristics are outside a threshold similarity with respect to one or more second audio characteristics previously stored in association with the user profile; and 
 based at least partly on the request for the operation and the one or more first audio characteristics being outside the threshold similarity with respect to the one or more second audio characteristics, refraining from performing the operation. 
 
     
     
       16. The method as recited in  claim 15 , further comprising:
 determining an audio signature based at least partly on the one or more second audio characteristics; and 
 determining that the audio signal corresponds to a second user different than a user associated with the user profile based at least partly on the audio signature. 
 
     
     
       17. The method as recited in  claim 16 , further comprising determining an identity of a user associated with the user profile based at least partly on the one or more second audio characteristics, the one or more second audio characteristics including a frequency associated with the audio signal. 
     
     
       18. The method as recited in  claim 15 , further comprising:
 receiving a second audio signal representing a second utterance, the second utterance being captured by the one or more microphones of the user device; 
 determining the one or more second audio characteristics associated with the second audio signal; and 
 storing the one or more second audio characteristics in association with the user profile. 
 
     
     
       19. The method as recited in  claim 15 , further comprising:
 causing, via one or more speakers of the user device, output of audio data indicating that the operation is not to be performed. 
 
     
     
       20. The method as recited in  claim 15 , further comprising requesting that a user associated with the user profile audibly output one or more second words included within a text message sent to the user device.